BACKGROUND: In this study, we sought to compare the long-term safety and efficacy of biphasic insulin aspart 30 (BIAsp30) with that of biphasic human insulin 30 (BHI30) over a period of 24 months in patients with type 2 diabetes. METHODS: Patients with type 2 diabetes (n=125) were assigned to twice-daily BIAsp30 or BHI30 and participated in both a 3-month initial period and a 21-month extension of a randomized, controlled, multinational trial. RESULTS: No significant difference was found in mean HbA(1c) after 24 months [BIAsp30, 8.35+/-0.20%; BHI30 8.13+/-0.16%; adjusted mean difference (BIAsp30-BHI30) 0.03 (90% CI -0.29 to 0.34)%, P=0.89]. The proportion of patients experiencing major hypoglycaemia was also similar during the first year (BIAsp30, 5%; BHI30, 8%; P=0.72), but it was significantly lower with BIAsp30 than with BHI30 during the second year (BIAsp30, 0%; BHI30, 10%; P=0.04). The proportion experiencing minor hypoglycaemia was not significantly different. No significant difference was recorded in changes in nonspecific insulin antibody levels after 24 months (BIAsp30, 4.87+/-1.92%; BHI30; 1.00+/-1.66%; P=0.13). Body weight change was 0.05+/-0.81 kg in the BIAsp30 group and 2.00+/-0.69 kg in the BHI30 group (P=0.07). CONCLUSIONS: Reduced major hypoglycaemia compared with BHI30 during the second year of treatment and comparable HbA(1c) levels after 24 months appear to support the hypothesis that the improved pharmacokinetic profile of BIAsp30 may favourably affect the balance between hypoglycaemia and hyperglycaemia in insulin-treated type 2 diabetes.